Probabilistic Fatigue Life Prediction Framework for Natural Rubber Considering Ambient Temperatures

ABSTRACT This paper focuses on the impacts of ambient temperature on the variability of the fatigue life of natural rubber (NR). Traditional physical models, such as those based on crack initiation or propagation, are limited to predicting fatigue life at a single ambient temperature. To overcome this limitation, a data‐driven model was developed to estimate the probabilistic fatigue life of NR while incorporating the influence of mechanical loads and ambient temperature. This model uses engineering strain peak, ambient temperature, and reliability as input parameters, with the probabilistic fatigue life as the output. The accuracies of two fatigue life prediction models were compared using the measured fatigue life data of NR under various temperatures. The results showed that the proposed data‐driven model achieved a higher prediction accuracy, with the predicted probabilistic fatigue life values falling within 1.5 times the deviation of the measured values across all tested ambient temperatures. These findings demonstrate the effectiveness of the proposed approach for accurately predicting the probabilistic fatigue life of NR.